Abstract:
The rapid growth in the amount of data generated from connected devices in smart industry, opens up new opportunities for improving the quality of service through data sh...Show MoreMetadata
Abstract:
The rapid growth in the amount of data generated from connected devices in smart industry, opens up new opportunities for improving the quality of service through data sharing. However, the key barriers for data sharing by the users of smart industries are security and privacy concerns. In addition to financial losses, the disclosure of sensitive information might cause major problems such as reputation damage, loss of competitive advantage, legal and regulatory violations, and even harm to individuals or society as a whole. In this paper, we propose FLEC, a federated learning framework for Cloud based smart industries. We use blockchain technology as an underlying architecture for data management and sharing between edge and cloud devices. Federated learning solves the problem of privacy leakage by sharing the model updates instead of raw data while blockchain provides a secure decentralized platform, robustness against poisoning attacks, and incentives to the participants. Finally, to make our approach well-suited for real-world non-iid data scenarios, we use batch normalization to alleviate the data heterogeneity problem. Extensive evaluation and assessment results demonstrate the superior performance of FLEC in aspects of accuracy, efficiency, and privacy protection.
Date of Conference: 28 May 2023 - 01 June 2023
Date Added to IEEE Xplore: 23 October 2023
ISBN Information: